Previous work by the Department revealed that surrogates' choices of which treatments patients should receive in a range of circumstances are often inconsistent with the patients own treatment preferences (Shalowitz 2006). Research in social psychology suggests that surrogate inaccuracy is just one example of a more general inability to predict the preferences and values of loved ones. This research finds that we all are subject to psychological biases in close relationships that are extremely difficult to correct. These biases suggest that improvements in surrogate accuracy are unlikely, a conclusion which is supported by our data showing that the two most widely-discussed methods for improvement appointing a surrogate, and discussing ones treatment preferences and values with this person are generally ineffective. To evaluate how patients want to have decisions made for them, once they recognize the extent of surrogate inaccuracy, the Department developed and conducted one of the largest surveys on patients preferences regarding surrogate decision making. This survey, which enrolled 1180 patients at George Washington Hospital and Clinics, was recently completed and data analysis has begun. This survey also was designed to evaluate patient attitudes toward using a patient preference predictor. Data analysis is in progress.